CN109884530A - A kind of battery management system and its operating method based on neural network algorithm - Google Patents
A kind of battery management system and its operating method based on neural network algorithm Download PDFInfo
- Publication number
- CN109884530A CN109884530A CN201910261973.3A CN201910261973A CN109884530A CN 109884530 A CN109884530 A CN 109884530A CN 201910261973 A CN201910261973 A CN 201910261973A CN 109884530 A CN109884530 A CN 109884530A
- Authority
- CN
- China
- Prior art keywords
- control module
- battery
- layer
- power monitoring
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The invention discloses a kind of battery management system and its operating method based on neural network algorithm, including power monitoring system main control module, charger, master controller, external system controller, from control module and ipc monitor module, the power monitoring system main control module passes through high-speed CAN line respectively and connect with master controller, charger, external system controller, power monitoring system main control module passes through electric current, Voltage On state in conducting wire and circuit, it is connect, is provided with from control module multiple with from control module by internal CAN line；The Monitor Computer Control System is provided with wireless transport module, is connect by wireless signal with power monitoring system main control module；The present invention improves SOC estimation precision in battery, can be monitored charge and discharge process in control, and can remotely be monitored by upper computer software.
Description
Technical field
The present invention relates to battery management system, specially a kind of battery management system and its behaviour based on neural network algorithm
Make method.
Background technique
SOC, full name are State of Charge, and battery charge state is also remaining capacity, and representative is that battery uses
A period of time or after lying idle for a long time it is remaining can discharge electricity amount and its fully charged state electricity ratio, commonly use percentage
Number indicates.
Traditional SOC estimation method: ampere-hour method, open circuit voltage method, impedance track method, three kinds of evaluation methods were actually using
There is the case where measurement inaccuracy, influenced by battery material, it is difficult to the accurate Estimation and Measurement of SOC in Cheng Zhong.
Summary of the invention
The purpose of the present invention is to provide a kind of battery management system and its operating method based on neural network algorithm, with
Solve the problems mentioned above in the background art.
To achieve the above object, the invention provides the following technical scheme: a kind of battery management based on neural network algorithm
System, including power monitoring system main control module, master controller, charger, external system controller, from control module and on
Position machine monitoring module, the power monitoring system main control module pass through respectively high-speed CAN line and master controller, charger,
External system controller connection, power monitoring system main control module pass through electric current, Voltage On state in conducting wire and battery circuit,
It is connect, is provided with from control module multiple with from control module by internal CAN line；The ipc monitor module is provided with wireless biography
Defeated module is connect by wireless signal with power monitoring system main control module.
Preferably, the master controller has been internally integrated RBF neural module, and the RBF neural module includes
Input layer, hidden layer, output layer are provided with both of which, respectively mode of learning and error propagation mode.
Preferably, the input layer, hidden layer, output layer be connected with each other, carry out data transmission, hidden layer, output layer it
Between data transmission it is in a linear relationship.
Preferably, the mode of learning is the detection ginseng that the input layer receives power monitoring system main control module
Number: electric current, voltage and temperature are normalized, are transferred in the neuron of hidden layer, carry out data change by hidden layer
It is transmitted to output layer after changing, SOC value is exported by output layer.
Preferably, when the error propagation mode is not inconsistent for the real output value of input layer with desired SOC value, pass through output
Layer carries out backpropagation to hidden layer, input layer, and training object is every layer of weight.
Preferably, in the RBF neural module, hidden layer is used as letter using input pattern at a distance from center vector
Several independents variable, and use radial basis function as activation primitive.
The present invention also provides a kind of operating methods of battery management system based on neural network algorithm, including walk as follows
It is rapid:
S1: the battery basic parameter of power monitoring system main control module connection, including battery cutoff voltage, capacity are determined
Deng；.
S2: the initial SOC of battery is determined；
S3: carrying out charge and discharge cycles experiment to battery, measures cell voltage during this, electric current, capacity, efficiency for charge-discharge
Situation of change；
S4: using the data measured in battery initial SOC and step S, RBF neural is trained, it is pre- to establish battery SOC
Survey model；
S5: experimental verification is carried out to known models, evaluated error repeats step S, continues to train, and be modified error.
Compared with prior art, the beneficial effects of the present invention are: the present invention passes through setting power monitoring system master control
Module is monitored battery；RBF neural module is set in the host controller, is received and is joined by RBF neural module
Number, estimates SOC；SOC estimation precision in battery is improved, charge and discharge process can be monitored in control, and can be with
It is remotely monitored by upper computer software.
Detailed description of the invention
Fig. 1 is the structural diagram of the present invention；
Figure label: 1, power monitoring system main control module；2, master controller；3, charger；4, external system controller；
5, from control module；6, ipc monitor module.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Referring to Fig. 1, the present invention provides a kind of technical solution: a kind of battery management system based on neural network algorithm and
Its operating method, including power monitoring system main control module 1, master controller 2, charger 3, external system controller 4, from
Module 5 and ipc monitor module 6 are controlled, the power monitoring system main control module 1 passes through high-speed CAN line and master control respectively
Device 2 processed, charger 3, external system controller 4 connect, and power monitoring system main control module 1 passes through conducting wire and battery circuit
Middle electric current, Voltage On state are connect with from control module 5 by internal CAN line, are provided with from control module 5 multiple；The host computer prison
Control module 6 is provided with wireless transport module, is connect by wireless signal with power monitoring system main control module 1.
Further, the master controller 2 has been internally integrated RBF neural module, the RBF neural module packet
Input layer, hidden layer, output layer are included, both of which, respectively mode of learning and error propagation mode are provided with.
Further, the input layer, hidden layer, output layer are connected with each other, and are carried out data transmission, hidden layer, output layer
Between data transmission it is in a linear relationship.
Further, the mode of learning is the detection that the input layer receives power monitoring system main control module 1
Parameter: electric current, voltage and temperature are normalized, are transferred in the neuron of hidden layer, carry out data by hidden layer
It is transmitted to output layer after transformation, SOC value is exported by output layer.
Further, when the error propagation mode is not inconsistent for the real output value of input layer with desired SOC value, by defeated
Layer carries out backpropagation to hidden layer, input layer out, and training object is every layer of weight.
Further, in the RBF neural module, hidden layer uses input pattern conduct at a distance from center vector
Argument of function, and use radial basis function as activation primitive.
The present invention also provides a kind of battery management system and its operating method based on neural network algorithm, including it is as follows
Step:
S1: the battery basic parameter that power monitoring system main control module 1 connects, including battery cutoff voltage, capacity are determined
Deng；
S2: the initial SOC of battery is determined；
S3: carrying out charge and discharge cycles experiment to battery, measures cell voltage during this, electric current, capacity, efficiency for charge-discharge
Situation of change；
S4: using the data measured in battery initial SOC and step S3, RBF neural is trained, battery SOC is established
Prediction model；
S5: experimental verification is carried out to known models, evaluated error repeats step S4, continues to train, and be modified error.
Working principle: it is connect by power monitoring system main control module 1 with battery circuit, the electricity in detection circuit
Stream, voltage, while temperature sensor is set, battery temperature is detected；Power monitoring system main control module 1 passes through height respectively
Fast CAN line is connect with master controller 2, charger 3, external system controller 4, is provided with RBF neural in master controller 2
Module；Charger 3 charges in battery capacity deficiency, and external system controller 4 controls external system, with master controller 2
Cooperating；Power monitoring system main control module 1 is connect by internal CAN line with from control module 5, is controlled each from control mould
Block 5 works；Meanwhile power monitoring system main control module 1 passes through wireless signal and ipc monitor module 6, by upper
The monitoring of machine remote software.
RBF neural module in master controller 2 is provided with input layer, hidden layer, output layer, hidden layer, output layer
Between data transmission it is in a linear relationship；" base " for using RBF as hidden unit constitutes implicit sheaf space, can will thus input
Vector maps directly to implicit sheaf space, without passing through power connection.After the central point of RBF determines, this mapping is closed
System also determines that.And the mapping in implicit sheaf space to output space be it is linear, i.e. the output of network is that hidden unit exports
Linear weighted function and, power herein is network tunable parameter.The effect of hidden layer is that vector is mapped to higher-dimension from the p of low dimensional
The case where h of degree, such low dimensional linearly inseparable, can become linear separability to high-dimensional, mainly be exactly kernel function
Thought.In this way, network is nonlinear by the mapping for being input to output, and network output for adjustable parameter but is linear
's.The power of network can directly be solved by system of linear equations, to greatly speed up pace of learning and avoid local minimum problem.
The hidden node of RBF neural is used as argument of function using input pattern at a distance from center vector, and uses radial base
Function is as activation primitive.The input of neuron is remoter from radial basis function center, and the activation degree of neuron is lower.
In practical operation step, the battery that power monitoring system main control module 1 connects is determined in step S1, S2
Basic parameter and determine the initial SOC of battery, in step s3, to battery carry out charge and discharge cycles experiment, measurement cell voltage,
The situation of change of electric current, capacity, efficiency for charge-discharge is trained study by RBF neural in step s 4, establishes battery
SOC prediction model carries out experimental verification to known models in step s 5, and evaluated error repeats step S4, continues to train, and
Error is modified.
There are two mode, respectively mode of learning and error propagation mode, the two process is opposite for RBF neural setting；
Mode of learning receives parameter by input layer, is normalized, is transferred in the neuron of hidden layer, is carried out by hidden layer
It is transmitted to output layer after data transformation, SOC value is exported by output layer；It is reversed to learn when the real output value and desired SOC value of input layer
The data procedures of habit mode, training object are every layer of weights.The training speed of RBF neural is fast, avoids local minimum,
It is connected to the network weight and output is in a linear relationship.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
A variety of variations, modification, replacement can be carried out to these embodiments without departing from the principles and spirit of the present invention by understanding
And modification, the scope of the present invention is defined by the appended.
Claims (7)
1. a kind of battery management system based on neural network algorithm, it is characterised in that: including power monitoring system master control
Module (1), master controller (2), charger (3), external system controller (4), from control module (5) and ipc monitor module
(6), the power monitoring system main control module (1) respectively by high-speed CAN line and master controller (2), charger (3),
External system controller (4) connection, power monitoring system main control module (1) pass through electric current, electricity in conducting wire and battery circuit
Crimping is logical, is connect, is provided with from control module (5) multiple with from control module (5) by internal CAN line；The ipc monitor mould
Block (6) is provided with wireless transport module, is connect by wireless signal with power monitoring system main control module (1).
2. a kind of battery management system based on neural network algorithm according to claim 1, it is characterised in that: the master
Controller (2) has been internally integrated RBF neural module, and the RBF neural module includes input layer, hidden layer, output
Layer, is provided with both of which, respectively mode of learning and error propagation mode.
3. a kind of battery management system based on neural network algorithm according to claim 2, it is characterised in that: described defeated
Enter layer, hidden layer, output layer to be connected with each other, carry out data transmission, and the linear pass of data transmission between hidden layer, output layer
System.
4. a kind of battery management system based on neural network algorithm according to claim 2, it is characterised in that:
Habit mode is the detection parameters that the input layer receives power monitoring system main control module (1): electric current, voltage and temperature
Degree, is normalized, is transferred in the neuron of hidden layer, is transmitted to output layer after carrying out data transformation by hidden layer, by
Output layer exports SOC value.
5. a kind of battery management system based on neural network algorithm according to claim 2, it is characterised in that: the mistake
When poor transfer mode is not inconsistent for the real output value of input layer with desired SOC value, carried out by output layer to hidden layer, input layer
Backpropagation, training object are every layer of weights.
6. a kind of battery management system based on neural network algorithm according to claim 2, it is characterised in that: described
In RBF neural module, hidden layer is used as argument of function using input pattern at a distance from center vector, and uses diameter
To basic function as activation primitive.
7. any one of a kind of operating method, including claim 1-6 of the battery management system based on neural network algorithm
A kind of battery management system based on neural network algorithm, characterized by the following steps:
S1: the battery basic parameter of power monitoring system main control module (1) connection, including battery cutoff voltage, appearance are determined
Amount etc.；
S2: the initial SOC of battery is determined；
S3: carrying out charge and discharge cycles experiment to battery, measures cell voltage during this, electric current, capacity, efficiency for charge-discharge
Situation of change；
S4: using the data measured in battery initial SOC and step S3, RBF neural is trained, battery SOC is established
Prediction model；
S5: experimental verification is carried out to known models, evaluated error repeats step S4, continues to train, and be modified error.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910261973.3A CN109884530A (en) | 2019-04-02 | 2019-04-02 | A kind of battery management system and its operating method based on neural network algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910261973.3A CN109884530A (en) | 2019-04-02 | 2019-04-02 | A kind of battery management system and its operating method based on neural network algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109884530A true CN109884530A (en) | 2019-06-14 |
Family
ID=66935782
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910261973.3A Pending CN109884530A (en) | 2019-04-02 | 2019-04-02 | A kind of battery management system and its operating method based on neural network algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109884530A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110534981A (en) * | 2019-08-30 | 2019-12-03 | 吉林大学 | A kind of anti-overcharge jack system of intelligence based on Internet of Things and its control method |
| CN112648923A (en) * | 2020-12-12 | 2021-04-13 | 山东省农业机械科学研究院 | Virtual calculation method for detecting size of product part |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102680903A (en) * | 2012-05-11 | 2012-09-19 | 山东轻工业学院 | Portable storage battery state detection system and method |
| CN105301508A (en) * | 2015-11-09 | 2016-02-03 | 华晨汽车集团控股有限公司 | Prediction method for electric automobile endurance mileage through redial basis function neural network |
| CN105823989A (en) * | 2016-05-04 | 2016-08-03 | 安徽工程大学 | Improved battery SOC prediction method for RBF neural network |
| CN107037373A (en) * | 2017-05-03 | 2017-08-11 | 广西大学 | Battery dump energy Forecasting Methodology based on neutral net |
-
2019
- 2019-04-02 CN CN201910261973.3A patent/CN109884530A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102680903A (en) * | 2012-05-11 | 2012-09-19 | 山东轻工业学院 | Portable storage battery state detection system and method |
| CN105301508A (en) * | 2015-11-09 | 2016-02-03 | 华晨汽车集团控股有限公司 | Prediction method for electric automobile endurance mileage through redial basis function neural network |
| CN105823989A (en) * | 2016-05-04 | 2016-08-03 | 安徽工程大学 | Improved battery SOC prediction method for RBF neural network |
| CN107037373A (en) * | 2017-05-03 | 2017-08-11 | 广西大学 | Battery dump energy Forecasting Methodology based on neutral net |
Non-Patent Citations (1)
| Title |
|---|
| 于智龙: ""基于自放电技术的电动车用锂动力电池SOC预测算法研究"", 《中国博士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110534981A (en) * | 2019-08-30 | 2019-12-03 | 吉林大学 | A kind of anti-overcharge jack system of intelligence based on Internet of Things and its control method |
| CN112648923A (en) * | 2020-12-12 | 2021-04-13 | 山东省农业机械科学研究院 | Virtual calculation method for detecting size of product part |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107367693B (en) | SOC detection system for power battery of electric vehicle | |
| CN107422272B (en) | Intelligent detection device for SOC of power battery of electric automobile | |
| CN112331941B (en) | Cloud auxiliary battery management system and method | |
| CN103675686B (en) | Electric vehicle power battery charging and discharging operating condition simulation system and method | |
| CN109407007A (en) | A kind of battery charge state detection method and device | |
| CN107176043A (en) | Power capacity for vehicle battery system is estimated | |
| CN109884530A (en) | A kind of battery management system and its operating method based on neural network algorithm | |
| CN106505557A (en) | A kind of remote measurement misidentification method and device | |
| CN107843845A (en) | A kind of power lithium-ion battery SOH estimation on line methods | |
| CN108001261A (en) | Power battery charged state computational methods and monitoring device based on fuzzy algorithmic approach | |
| CN102967831B (en) | On-line detection system and detection method of lead-acid storage battery performance | |
| CN104821632A (en) | Battery system charging voltage balance control method and system | |
| CN106998086A (en) | MW class energy-accumulating power station battery management method and its system | |
| CN104142446B (en) | System and method for testing charging control response time of electric vehicle charger | |
| CN112714156A (en) | Cloud BMS (battery management system) cooperative management method and system and vehicle | |
| CN204142849U (en) | For simulating battery simulation device and the system of charging electric vehicle | |
| TWM631494U (en) | Intelligent Power Management Edge Estimation System | |
| CN113985296A (en) | Hardware-in-loop test system for battery management system | |
| CN107688153B (en) | Distributed wireless bus battery screening and testing system and testing method | |
| CN206542219U (en) | A kind of MW class energy-accumulating power station battery management system | |
| CN107546809A (en) | It is a kind of based on strain and the intelligent charging system and method for temperature and regulating and controlling voltage | |
| CN104730938B (en) | A kind of battery characteristics analogy method and system | |
| CN115932592A (en) | Electric vehicle battery fault diagnosis method based on digital twin model | |
| CN116031508A (en) | Lithium ion battery management system and method based on deep learning | |
| CN115101836A (en) | Cloud-end converged battery system management method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190614 |
|
| RJ01 | Rejection of invention patent application after publication |